Working Paper No.

Sustainable Economic Growth

and Exhaustible Resources:
A Model and Estimation for the U.S.

by

Almuth Scholl and Willi Semmler

University of Bielefeld
Department of Economics
Center for Empirical Macroeconomics
P.O. Box 100 131
33501 Bielefeld, Germany
Sustainable Economic Growth and Exhaustible Resources: A
Model and Estimation for the U.S.y

Almuth Schollz and Willi Semmlerx

January 2000

Abstract
This paper studies current models on sustainable economic growth with resource
constraints and explores to what extent resource constraints can be overcome by
substitution and technological change. We also study the problem of intergenera-
tional equity and the dierent criteria that have been suggested in the literature.
The central part of this paper is the presentation of stylized facts on exhaustible
resources and an estimation of a basic a model with resource constraints for U.S.
time series data. The estimated years left until depletion and the empirical trends
of the ratios of capital stock and consumption to resources seem to indicate that
there might be a threat to sustainable growth in the future. In our estimation we
obtain parameter values which help to interpret the extent to which growth with
exhaustible resources is sustainable.

y
We would like to thank Toichiro Asada for helpful comments
z
Dept. of Economics, Humboldt University, Berlin
x Dept. of Economics, University of Bielefeld, Germany and New School for Social Research, New York
1 Introduction
Since the early 1970's there is growing concern that human activity depletes and pol-
lutes the environment. As economic decisions are restricted by the niteness of natural
resources and by the limited capacity of the nature to absorb pollution, attention is de-
voted to the question whether it is possible and desirable to continue present patterns of
economic growth.
In 1972 economists like Meadows et al [40] or Daly [15] formulated pessimistic predictions
about a "sudden and uncontrolable decline in both population and industrial capacity" if
no "conditions for ecological and economic stability that is sustainable far into the future"
are established1. Other economists like Beckerman [5] have the optimistic view that tech-
nological progress and the discovery of new substitutes make continued economic growth
possible. A general consensus of the economic growth debate is that there are trade-os
among environmental and economic goals. The agreement is that economic activity which
ignores the biological or social system is not sustainable. There exist many dierent de-
nitions of sustainability but all of them have two points in common. First, they recognize
that resource and environmental constraints aect the patterns of development and con-
sumption in the long-run. Second, they are concerned about equity between generations
(intergenerational equity). One of the most famous denitions is stated by the Brundtland
Commission [74] in 1987: "Sustainable development is development that meets the needs
of the present without compromising the ability of future generations to meet their own
needs". Similarly Solow [62] denes sustainablity as "an obligation to conduct ourselves
so that we leave to future the option or the capacity to be as well o as we are." Pearce,
Barbier and Markandya [48] point out that "natural capital stock should not decrease over
time" whereas Pezzey [50] denes sustainable economic growth as "non-declining output
or consumption over time" and sustainable economic development as "non-declining util-
ity over time".
As to the sustainability of natural resources one can distinguish between renewable and
non-renewable resources. Dynamic models on renewable resources can be found in Clark
[12], Sieveking and Semmler [58] and Semmler and Sieveking [56]. In those papers theo-
rems on the sustainability of resource economics which have been developed by studying
one resource only are evaluated for the case when resources interact as an ecological sys-
tem. For the latter case resource management problems and policies aiming at conserving
resources are studied. Furthermore, in Semmler and Sieveking [57] credit nanced extrac-
tions of resources are considered and the fate of resources explored.
The current paper deals with dynamic models with exhaustible resources. We discuss
prototype growth models that incorporate and study the consequences of nitely avail-
able exhaustible resources. Some of the problems studied here will also arise in the case
of renewable resources so for example, the problem of intergenerational justice. The re-
1 Meadows [40], p. 23.

1
mainder of the paper is organized as follows. In section 2 we survey growth models with
natural resource constraints. Section 3 discusses the problem of intergenerational justice.
Section 4 presents stylized facts on exhaustible resources pertaining to the U.S.. Section
5 presents the estimation of our growth model. Section 6 concludes the paper.

2 Modern Growth Models and Resource Constraints

We here review growth models which take into account the presence of exhaustible natural
resources. The main interest of the analysis is the question to what extent the process of
economic growth is restricted by the niteness of resource stocks and whether sustained
consumption and utility levels are feasible.
We rst consider a basic model where the exhaustible natural resource is used as an input
for the production of a good which is then either consumed or added to the capital stock
to enhance future production. Then, the consequences of dierent extensions concerning
technological progress are analyzed, and, nally, it is assumed that natural resources may
provide services in preserved states.
Before turning to the description of the models it is important to have a clear distinction
between renewable and exhaustible resources. The main feature of an exhaustible resource
is that its growth rate is nil, and that it is unrecycable. Furthermore, it is used up when
used as an input in production. To study a meaningful economic problem, the natural
resource must be essential, i.e. production is impossible without it.2
2.1 The Basic Model

Dasgupta and Heal [18], Stiglitz [65] and Solow [61] analyze the optimal depletion of
exhaustible natural resources in the context of a growth model where the resource is used
as an input for the production of a composite commodity. The production function F
depends on the ow of the exhaustible resource at date t and on the stock of a reproducible
good at date t. In order to obtain greatest possible social welfare the present value of
utility U derived from consumption Ct of the produced good is maximized subject to
the evolution of the reproducible captital stock Kt and the constraints imposed by the
niteness of the resource stock St :
R
Max 01 U (Ct )e Æt dt (1)
s:t:
K_ t = F (KtZ; Rt ) Ct
1
St = S0 Rt dt
0
S_ t = Rt
2 Dasgupta and Heal [19].

2
Æ denotes the discount rate, and Rt is the ow of the exhaustible resource. The ini-
tial capital stocks K0 and S0 are strictly positive and given. The production function
F (Kt ; Rt ) is assumed to be increasing, strictly concave, twice continuously dierentiable
and homogenous of degree unity. The utility function U (Ct ) is supposed to be strictly
concave, and for Ct ! 0 its rst derivative is innity. Here, the extraction of the resource
is assumed to be costless.
Solving the maximization problem and combining the optimality conditions yields the
following results (for details see appendix A.1): First, along an optimal path the rate
of consumption depends on the discount rate Æ, on the elasticity of marginal utility of
consumption  and on the marginal productivity of reproducible capital FK :
C_ t FK Æ
C
=  (2)
t

with (Ct ) = CUt U(C(Ct )t ) and FK = @F (@K

00

t . Equation (2) states that the higher the

Kt ;Rt )

discount rate the more the rate of consumption falls over time along an optimal path.
0

Second, along an optimal path the rates of return of exhaustible and reproducible capital
are equal:
@F 1
F = RK (3)
@t FR
with FR = @F (@RKtt;Rt) . If the production function is homogenous of degree one, it is possible
to set xt = KRtt with f (xt) = F ( KRtt ; 1). Substituting FR = f (xt ) xt f 0(xt ) and FK = f 0(xt )
in equation (3) yields the following capital-resource ratio along an optimal path:
x_t
xt
=  f (xt)
xt
(4)
with
f 0 (xt )(f (xt ) xt f 0 (xt ))
=
xt f (xt )f 00 (xt )
as the elasticity of substitution between reproducible capital and the exhaustible resource.
Equation (4) represents the rate at which reproducible capital is substituted for the ex-
haustible resource. It depends on the elasticity of substitution and on the average product
per unit of xed capital.
In order to conclude whether a positive level of consumption is sustainable over time Das-
gupta and Heal [18] analyze an economy where output is produced by a CES - production
function, i.e. the case of a constant elasticity of substitution. There are three cases to
mention:
1.  = 1 (i.e. the Cobb-Douglas-production function)
The exhaustible resource is essential and innitely valuable at the margin, whereas
the asymptotic value of marginal productivity of capital is zero. Solow [61] concludes
3
 that sustained per capita consumption is feasible if the share of capital exceeds that
of natural resources.
2. 0   < 1
The exhaustible resource is essential but nitely valuable at the margin. Thus, a
positive and nondecreasing level of consumption over an innite time horizon is not
sustainable.
3. 1 >  > 1
Sustained consumption is feasible because in this case the exhaustible resource is
inessential.
2.2 Technology

The basic model can be augmented by introducing technical change which makes it eas-
ier to nd new substitutes in order to render an essential natural resource inessential.
Dasgupta and Heal [18] assume that technical progress is uncertain: the exact date of
discovering a substitute and its detailed characteristics and usefullness are unknown. The
new technique is supposed to occur at an unknown date T which is a random number
with an exogenously given probability density function !t:
P robabilityZ (T = t) = !t
1
!t dt = 1
0
!t > 0
In order to express the situation of uncertainty, the objective is to maximize the expected
present value of utility. After some manipulation one obtains the following maximization
problem:3
R
Max 01 [U (Ct )
t + !t W (Kt ; St )]e Æt dt (5)
s:t:
K_ t
= F (KtZ; Rt) Ct
1
St = S0 Rt dt
0
S_t = Rt
with
t = Rt1 !t dt and W (Kt ; St) = Max RT1 U (Ct )e Æ[t T ] dt.
Kt , Ct , Rt and St are all non-negative and the initial values K0 and S0 are given.
Solving the maximization problem, combining the rst order conditions, and arguing that
3 Dasgupta and Heal [17], pp.20.

4
at the discovery date of the substitute the then existing stocks of reproducible and natural
capital have no economic value anymore because the new technology is more eÆcient (i.e.
WK = WS = 0), allows for the following conclusions (for details, see appendix A.2): First,
along an optimal path the rate of consumption depends on a modied discount rate:
C_ t FK (Æ + t )
= Ct 
(6)
with t =
!tt as the conditional probability of the technological breakthrough at date t
given the substitute has not been discovered earlier. The discount rate is modied by
the addition of the factor t showing the probability of the essential resource becoming
inessential as a result of technical progress. Thus, in a situation of uncertainty the discount
rate is higher than in a situation of certainty.4 Obviously, the equation describing the
ratio of capital-resource input is the same as before:
x_t
=  f (xt)
xt xt
(7)
Stiglitz [65] examines an economy where output is produced by a Cobb-Douglas-production
function. He concludes that sustained per capita consumption is feasible, if there is a re-
source augmenting technical change at any positive rate (for   1). Toman, Pezzey and
Krautkraemer [67] point out that for the case  < 1 sustained per capita consumption is
possible if technological progress is high enough.
2.3 Backstop Technology

So far it is assumed that the natural resource is exhaustible, i.e. once it is used up, it
is impossible to nd more, and that the extraction is costless. As an extension it is now
supposed that the resource is available in unlimited quantities, but at various grades and
various costs. For example the ores of a number of metals can be extracted from the
deposits currently used which are exhaustible. If they are used up, the metals themselves
can be extracted from the sea or from rockformations, which is much more expensive.
Thus, at higher prices the natural resource may be of unlimited availability. Heal [31]
calls this a backstop technology. We can incorporate it into the basic model described in
section 2.1. The total amount of the resource used at date t is denoted as follows:
Z 1
z (t) = Rt dt
0
It is assumed that at date T the conventional deposits are exhausted and a backstop
technology takes over. Up to the level zT the extraction costs rise with cumulative ex-
traction, then the backstop technology is available at a constant cost per unit b. g(zt )
4 See also Sieveking and Semmler [58]

5
denotes the extraction costs per unit at date t with @z@gt = g0(zt ) > 0 for 0  zt  zT
and g(zT ) = b > 0 for zt  zT .
The maximization problem is solved in two steps (see appendix A.3): First, the situation
is examined before current deposits are exhausted (maximization problem (8) ), second,
the situation is examined after the backstop technology has taken over (maximization
problem (9)).
R
Max 01 U (Ct )e Æt dt (8)
s:t:
K_ t
= F (KtZ; Rt ) Ct g (zt )Rt
1
St = S 0 Rt dt
0
S_t = Rt
where g(zt)Rt represents the total extraction costs.
R
Max 01 U (Ct )e Æt dt (9)
s:t:
K_ t = F (Kt ; Rt ) Ct bRt
The initial capital stocks K0 and S0 are strictly positive and given. Computing the
conditions along an optimal path of problem (8) yields:
C_ t FK Æ
Ct
=  (10)
and
@F 1 F g (z )
FK = R + K t (11)
@t FR FR
Substituting FK = f 0(xt ) and FR = f (xt) xt f 0(xt ) with f (xt ) = F ( KRtt ; 1) results in the
following capital-resource ratio along an optimal path:
x_t f (xt ) f 0 (xt ) g (zt )
xt
= xt
+ xt f 00 (xt ) xt
(12)
Condition (12) is a generalization of condition (4). Here, the equation is augmented by
the term xtff (x(tx)t ) g(xztt) which re ects the cumulative costs of extraction.
0

Following Heal [31] we can draw the following conclusions. During the initial period the
00

lower-cost stocks of the natural resource are exhausted, and the path of the economy is
described by problem (8) and conditions (10) and (11). The dierence between prices
and extraction costs, i.e. the user costs, decline until they reach zero at date T when the
backstop technology takes over because the lower-cost stocks are totally used up. From
then on, the economy behaves according to problem (9), thus, the extraction costs of the
natural resource always equal its price.
6
3 Intergenerational Equity
As apparent from the above models the depletion of resources generates externalities for
future generations. Therefore, the problem of intergenerational equity arises. In order to
study this problem we rst will show how natural resources may aect the welfare of the
society.
3.1 The Amenity Value of a Natural Resource

There are two ways of how a natural resource contributes to society's welfare. The models
described so far refer to the rst way: the resource is utilized as an input factor for the
production of a composite commodity. The second way of a natural resource serving for
the well-being of the society is that it may provide valuable services in preserved states,
that is scientic, recreational, and aesthetic values. To take into account these so-called
amenity values of natural resources the resource stock St is included in the utility function
(Krautkraemer [37]). The objective is to maximize present value of utility (see appendix
A.4):
R
Max 01 U (Ct ; St )e Æt dt (13)
s: t:
K_ t = F (KtZ; Rt) Ct
1
St = S0 Rt dt
0
S_t = Rt
where Ct , Kt , Rt , St are all non-negative. U (Ct ; St) is assumed to be twice continuously
dierentiable with UC = @U (@CCtt;St) > 0 and US = @U (@SCtt;St) > 0, UCC = @ U@C(Ctt ;St) < 0 and
2
2

USS = @ U@S(Ct ;St) < 0 and lim

Ct !0 UC = 1. The production function F (Kt ; Rt ) has the
2
t
same properties as before. Solving the maximization problem and combining the rst
order conditions yield the following conditions for the rate of consumption and for the
capital-resource ratio along an optimal path:
C_ t FK Æ (UCS =UC )Rt
Ct
= 
(14)
x_t
=  f (xxt ) + xU2 fS =U C
(15)
t (xt )
xt t
00
The amenity services of a natural resource aect the extraction and the consumption
mainly through their contribution to the rate of return to the resource stock. Resource
amenities enhance the value of the resource stock. Therefore, the initial price of the
7
resource must be higher than in the previous models. Furthermore, since f 00(xt ) < 0,
the amenity services lower the rate of change in the capital-resource ratio by the term
xt f (xt ) . The higher the marginal rate of substitution between amenities and consumption,
US =UC
2
the higher the reduction of the rate of change of the input ratio.
00

In this framework where natural environments are valued in their own rights the so-called
Green Golden Rule 5 can be introduced which is motivated by the Golden Rule of Economic
Growth 6 . The Golden Rule of Economic Growth gives the growth path with the highest
indenitely maintainable level of consumption, whereas the Green Golden Rule focuses
on the highest indenitely maintainable level of instantaneous utility. Thus, the Green
Golden Rule incorporates the aim of sustainability. Formally the rule can be written as:7
maxfeasible paths limt!1 U (Ct ; St )
If the resource is used as an input factor for the production of a composite commodity,
in the long-run the only constant level of resource input is zero. Since the resource
is essential, no output can be produced and consumption is zero. Hence, the highest
indenitely maintainable utility level is feasible if the total initial stock is conserved.
3.2 Intergenerational Equity

Standard growth models which incorporate the concept of sustainability focus on the con-
sequences of natural resource constraints on the long-run pattern of economic development
and consumption. In order to determine intertemporal welfare recent growth theory has
used the concept of discounted utilitarism, that is the future is discounted in comparison
with the present. Ramsey [52] states that discounting is "ethically indefensible and arises
merely from the weakness of the imagination" because a positive discount rate results
in an asymmetric treatment of present and future generations. Thus, discounted future
utility neglects intergenerational equity as the second important point of the concept of
sustainability. In the following, we want to give a brief review of some alternative concepts
which try to meet the objective of a fair treatment of dierent generations.
A simple way to account for intergenerational equity is to assume the case of a zero utility
discount rate, that is present and future generations are given the same weight. Another
alternative is to apply the "overtaking criterion" as proposed by Weizacker [73] which
states that one consumption path is better than another if from some date on total utility
of that path is greater. Formally, if
Z T Z T
U (C 1 ) dt 
t U (Ct2 ) dt
0 0
5 Chichilnisky [11] and Beltratti, Chichilnisky, Heal [10].
6 Phelps [51].
7 Heal [34], p. 43.

8
But applying these approaches gives rise to technical problems. For a zero discount rate
the set of attainable values of the integral may be open, and the way of ranking consump-
tion paths according to the overtaking criterion is incomplete.
According to the Rawlsian Criterion8 intergenerational equity means: Maximize the wel-
fare of the less advantaged generation. Formally,
maxfeasible paths mingenerations t (W elfaret )
The consequence of this decision rule is that the welfare level should be the same for
all generations. If a later generation enjoys higher welfare, an earlier generation should
increase its own welfare at the expense of the later generation and vice versa. Solow
[61] points out that in comparison with the discounted solution based on utilitarism the
Rawlsian Criterion will use up the natural resource stock faster. Since the utilitarian rule
demands higher savings, earlier generations will have a lower standard of living than the
constant max-min rule would generate. The Rawlsian Criterion has two main diÆculties:
First, a society needs an initial capital stock high enough to make a decent standard of
living possible, but the explanation of its existence is missing. Second, the rule does not
yield a reasonable result if ongoing technical progress is assumed.9
More recently, Chichilnisky [11] denes two axioms for sustainability which deal with the
problem of intergenerational equity. The rst axiom states that the present generation
should not dictate the outcome in disregard for the future. The second axiom states that
the long-run future should not dictate the present. Welfare criteria which do satisfy the
two acioms are called sustainable preferences. In order to formulate a criterion which does
belong to the class of sustainable preferences positive weight is placed on the present and
on the very long-run properties of a growth path. Formally,
Z 1
 U (Ct ; St )
(t) dt + (1 )limt!1 U (Ct ; St )
0
where  2 (0; 1).
(t) is any measure with R01
(t) dt = 1. If
(t) = e Æt , the rst term is
just the discounted integrals of utilities. The second term re ects the limiting properties of
the utility stream, and it already has been mentioned as the Green Golden Rule solution.
The Chichilnisky Criterion places more weight on the future than the standard approach
of discounting utility but less than the Green Golden Rule. It is possible to apply the
Chichilnisky Criterion to neoclassical growth models at the aforementioned type 10 but
nding the solution is quite complicated. We therefore leave aside detailed discussions.
This very short review of dierent welfare criteria has shown that it is very diÆcult to
nd approaches which do meet the objective of permitting intergenerational equity and
which are technically operable at the same time. For that reason, discounted utility is
8 Rawls [53].
9 A more elaborate version of the Rawls criterion is proposed in Semmler and Sieveking [58]
10 For a detailed analysis see Heal [33].

9
still dominant, as it is the technically most convincing approach, though it favours the
present over the future.

4 Stylized Facts
Economic theory states that substitution possibilities, technological progress and the value
of the resource in preserved states may prevent the total depletion of natural capital. In
this section the patterns of some selected non-renewable resources of the US economy
are analyzed from 1960 to 1995. This than serves as background for our estimations in
section 5. Here, extraction rates, available resource quantities today and in the future
are examined with respect to the question whether there are reasons to argue that the
resources will be soon exhausted or that improved technology and the development of
reproducible substitutes make a sustainable economic development possible.
Two dierent kinds of natural resources fulll the main characteristics of depletion: fuel
minerals such as energy resources, and non fuel minerals such as metals and industrial
minerals. When talking about the limited availability of natural resources, it is important
to have clear denitions of the dierent components of which the total resource stock
consists. Figure 1 explains the dierent components.

10
 Figure 1: The Components of Resource Stock
The reserves of the discovered resources consist of proved reserves and other reserves
such as inferred reserves (eld growth), measured reserves and indicated reserves. Proved
reserves are those amounts of the resource that geological and engineering data demon-
strate with reasonable certainty to be recoverable in the future from known reservoirs
under existing economic and technological conditions. The other reserves consist on the
one hand of that part of the identied economically recoverable resource that will be
added to proved reserves in the future through extensions, revisions and the discovery of
new elds in already discovered regions, and on the other hand on those quantities of the
resource that may become economically recoverable in the future from existing production
reservoirs through the application of currently available but as-yet uninstalled recovery
technology.11 For details on the data sources for the subsequent summaries, see appendix
B.
As noted before, fuel minerals are energy sources such as crude petroleum, coal and
natural gas. Since during the last thirty years the US economy has experienced continued
economic growth, that is a rising level of real GDP, total energy consumption has increased
11 For denitions see Energy Information Administration, (1996, 1997).

11
by roughly 35 percent. To satisfy increasing energy demand the production of especially
coal and natural gas have risen. In order to draw conclusions whether the resources are
used more eÆciently over time it is interesting to analyze the patterns of production rates
per dollar of real GDP, see Figure 2.

Figure 2: Fuel: mineral production per 1990 $ GDP (coal production measured short
tons, crude petroleum production measured in barrels, natural gas production measured
in millions of cubic feet.)
Table 1 summarizes the results.
Table 1: Fuel Mineral Production per dollar real GDP

crude coal in short natural gas in

petroleum tons in millions of
barrels cubic feet
1960 1.2680*10 3 2.0526*10 4 6.3155*10 6
1995 3.8447*10 4 1.6589* 10 4 3.1316*10 6
change in % -70 -19 -48
12
 For every energy resource the production rates per GDP are falling, i.e. crude petroleum,
coal and natural gas are used more eÆciently today which may be the result of improved
technology. Nonfuel minerals are on the one hand metals and on the other hand indus-
trial minerals. Here, only some selected metals such as copper, iron ore, lead and zinc are
analyzed. The production rates of copper, iron ore, lead and zinc behave quite unregulary
from 1960 to 1995 and no trends can be determined.

Figure 3: Metal: Production per GDP (metal production in short tons)

Figure 3 and Table 2 show that the production rates per dollar real GDP are decreas-
ing.

Table 2: Nonfuel Mineral Production per dollar real GDP

copper in iron ore in lead in zinc in

short tons short tons short tons short tons
1960 5.3287*10 4 4.0952*10 2 1.2187*10 4 2.1463*10 4
1995 2.9700*10 3 9.8124*10 3 6.1990*10 5 9.8605*10 5
change in % -44 -76 -49 -54
13
 Exhaustible energy resources can be substituted by renewable energy sources like for
example wind and sun power.

Figure 4: Energy production sources in percent of total

Figure 4 shows the shares of the dierent energy sources. Today, nuclear power and
renewable energy together are just 20 percent of total energy production. But the trend is
that the production of crude petroleum, coal and natural gas is decreasing while nuclear
and renewable energy production is increasing.
Copper, lead and zinc are metals which can be substituted by aluminium and plastic. For
iron ore there do not exist any substitutes but as it will be shown later the reserves of
iron ore are very high and will last for the next centuries.
In order to draw conclusions about future availability of exhaustible resources it is
necessary to have a closer look on the amount of proved and other reserves of a resource
in comparison with its production rates.

14
 Figure 5: Reserve-production ratio
For the exhaustible energy resources crude petroleum and natural gas it is possible to
plot the reserve-production ratio for the observed time period, see Figure5. The smaller
the ratio the scarcer is the natural resource. For petroleum and natural gas the trend is
declining, thus, proved reserves will be depleted soon if for example extensions/discoveries
of new elds in already discovered regions or new discoveries do not make reserve addi-
tions possible.
Every year the Energy Information Administration and the US Geological Survey esti-
mate quantities of technically recoverable resource amounts that could be added to the
the already proved reserves of the US. It is interesting to ask how many years it will take
to exhaust the today estimated technical recoverable resources quantities. As the pro-
duction rates of the energy resources coal and natural gas are steadily increasing during
the observed time period it is assumed that they continue to increase with an average
production growth rate. The other resources do not show any clear trend in their produc-
tion rates and, therefore, it is supposed that production will continue to follow a stable
pattern during the next years. Tables 3 to 5 summarize the results.

15
 Table 3: Estimated Reserves of Petroleum and Natural Gas 1995

crude petroleum in natural gas in trill.

bill bbl. cu. ft.
proved reserves 20.2 135
eld growth 60.0 322
undiscovered resources 30.3 259
total reserves 110.5 716
average annual production /
average production growth rate 3.0076 0.012
years left until exhaustion 40 300

Table 4: Estimated Reserves of Coal 1996

proved other total average production years left

reserves reserves total reserves growth rate until exhaustion
19 428 507 740 527 168 0.026 250

Table 5: Estimated Reserves of Some Selected Metals 1997

copper in iron ore in lead in 1000 zinc in 1000

1000 metric 1000 metric metric tons metric tons
tons tons
proved reserves 45 000 10 000 000 6 500 25 000
other reserves 90 000 23 000 000 20 000 80 000
total reserves 135 000 33 000 000 26 500 105 000
average
annual production 1414.25 68844.44 439.72 437.14
years left
until exhaustion 95 480 60 240
Although the results of the empirical analysis are restricted by the length of the ex-
amined time period and there is a clear decline and niteness of resources.
16
Yet, there may be technological progress that makes a further development of renewable
substitutes for exhaustible resources possible. As the US economy has experienced con-
tinued economic growth during the second half of this century the production rates of
exhaustible resources have risen and for all analyzed minerals the production per dollar
real GDP has declined. Increased eÆciency, as the consequence of advanced technology
and the use of close substitutes are likely explanations for that fact.
Assuming no changes in production trends yields predictions of the number of years left
until present estimated resource reserves are depleted. The estimated reserves of coal
and natural gas will last for the next two or three centuries. As the present estimated
reserves of crude petroleum will be depleted in 40 years the US economy is dependent
on oil imports mainly from Arab countries. Substitutes of coal, natural gas and crude
petroleum are nuclear and renewable power. Since the share of these alternative energy
production sources has increased from 1960 to 1995 it seems likely that further research
and development will enhance this trend. The reserves of iron ore and zinc are very high
while lead and copper seem to be rather scarce. Yet, an increased use of particularly plas-
tics as a renewable substitute may relax the constraints that a sustainable development
faces.
These predictions are all made under the assumption that only currently available tech-
nology is applied. It seems to be likely that technological progress improves for example
mining and rening methods or makes discoveries of new resources elds possible, and,
therefore, enhances present estimations of available reserves. In order to ensure sustain-
able growth in the sense of minimal degradation of natural capital stocks and intergener-
ational equity, it appears to be important to develop further renewable substitutes.

5 Estimation of the Model

In order to estimate the growth model as described in section 2, rst, consumer preferences
and the technology of producing goods have to be specied, and, second, a data set has to
be constructed. In our estimation we focus on the standard model as presented in section
2.1.
It is assumed that the natural resource contributes to economic activity only in one way:
it is used as an input factor for the production of a commodity which is either consumed
or added to the capital stock. The present value of utility received from consumption is
given by Z 1 1 
Ct 1 e Æt dt
0 1 
where, , is the elasticity of marginal utility. The technology of goods is described by
a Cobb-Douglas-production function which depends on reproducible capital Kt and the
exhaustible resource ow Rt . The elasticity of substitution between reproducible capital

17
and the exhaustible resource  equals 1. The evolution of capital is determined by
K_ t = Kt Rt1  Ct
 denotes the share of reproducible capital in production. Maximizing present value of
utility and setting yt = RCtt yields the following estimable system: For a detailed study of
the solution, see appendix A.5
y_t xt 1 Æ
yt
= 
 (16)
x_t
xt
= x 1
t (17)
R_ t
Rt
=  (18)
with  as the growth rate of the exhaustible resource ow.
As time series data we need consumption, reproducible capital stock12 and the exhaustible
resource ow.13 The reproducible capital stock, Kt, is gross real private xed capital
stock and Ct is private consumption. The time series for the resource, R, is based on our
own computation. Since the total mineral production value is the amount of extracted
exhaustible resources times average prices, it is used to measure the exhaustible resource
ow. All time series are de ated by the GDP-price index 1990=100.
Equations (16), (17) and (18) are estimated directly by using non-linear least squares
techniques (NLLS).14 In the estimation we have prexed the discount rate, Æ, and the
parameter of relative risk aversion, . The reason for this procedure is that the model
we are considering - leaving aside substitution, technological change and the role of other
inputs - is in its current form rather incomplete and reliable estimates for the discount
rate as well as for relative risk aversion cannot be expected. Therefore, we prex them
at levels that have been obtained by other recent studies (see Semmler and Gong, 1997).
The estimation results are summarized in table 5.
Table 6: Estimation Results

parameter value standard error

 0.32 3.4668
Æ 0.03 0.0369
 0.5 0.2488
 0.002 0.1856
12 Data on consumption and capital stock are obtained from Citibase (1998)
13 Source: US Department of Commerce, Economics and Statistics [68].
14 Computed with GAUSS - OPTMUM version 3.00.

18
 The estimated capital share in income,  , and the estimated growth rate of the re-
source ow, , are reasonable.
Although there is a very unregular behaviour of the total mineral production value over
the observed time period the data show an enormous increase of the value of resources in
the years 1972 to 1981 caused by the price eects of the oil crisis.

Figure 6: Capital-Resource Ratio

Figures 6 and 7 show the estimated (tted) and actual time series for the ratio of
capital stock to resources and the ratio of consumption to resources. Both gures show
that there is, in particular, since 1980, a strong trend toward the depletion of resources
relative to capital stock and consumption.

19
 Figure 7: Consumption-Resource Ratio
As the gures show the model with our estimated parameters matches the data well.
As already noted, the very simple structure of the model may explain the observable
slight correlation of the error terms. We also have supposed that population remains
constant which implies that labour as an input factor in production is constant, too. This
assumption seems justied when analyzing the very short-run, but not appropiate when
examining a time period of 35 years. It seems to be likely that incorporating the factor
labour into the model would improve the estimation results. Such an improvement is
particularly critical for our estimation result on  , since our estimate gives a value of
 = 0:32 This would mean, according to Solow [61], that sustained per capita growth is
not feasible (see section 2.1). Improvement of our estimations by including further factors
would most likely give us the result that exhaustible resource share in production, 1  , is
likely to be smaller. Furthermore, as the stylized facts support, it would be reasonable to
allow technological progress and substitution. Finally, the quality of nonlinear estimations
depends strongly on the number of obversations. Since the analyzed time period consist of
only 35 data points, it is diÆcult to achieve suÆcient robustness in the estimations. This
problem was clearly observable when we attempted to estimate the prexed parameters 
and Æ. Their estimation in fact turned out to be non-robust with respect to the algorithm
used. We therefore kept them prexed.
In summary, although our preliminary results may be illuminating future research should
20
take into account the factor labour and technological progress and substitution, and to
estimate such a model over a longer time period or with data with a higher frequency.

6 Conclusions
This paper attempts to study in a formal model the growing concern that human activity
and economic growth depletes natural resources. Although both, renewable as well as
exhaustible resources are threatened by extinction in the process of economic growth15 this
paper in particular focuses on exhaustible economic resources. We pursue the question
of whether the current rate of extraction of exhaustible resources is sustainable given
the present pattern of economic growth. We present time series data and give rough
estimates of depletion time for exhaustible resources. We also study the problem of
intergenerational equity and the dierent criteria that have been suggested in the literature
to conserve resources and do justice to future generations. In a particular growth model
we study to what extent resource constraints can be overcome by technological change
and substitution. Although it would be advisable to consider world-wide trends in the
exhaustion of resources we, because of data problems, restrain our study to empirical
trends and stylized facts on exhaustible resources of the U.S. economy. We estimate
a standard growth model with resource constraints for U.S. time series data. We can
observe from our time series data that there is, disrupted by the strong increase of the
value of resources due to the oil crisis in the 1970s, a strong depletion of resources relative
to capital stock and consumption since 1980. Our econometric estimations point into
the same direction. Yet, our estimates, in particular our estimate of the capital share,
which indicates { if one follows Solow [61]{ non-sustainable growth, has to be interpreted
with some care since the estimated growth model lacks other variables, uses limited time
series and low frequency data. Moreover, the lack of data on technological change and
substitutes for exhaustible resources prevents us from drawing the strong conclusion that
there is a threat to sustainable growth in the future. Yet, the ratio of capital stock or
consumption to resources has tripled possibly indicating future problems concerning the
availability of exhaustible resources.

15 Forstudies on the extinction of renewable resources, see Clark [12] and Semmler and Sieveking [1994,
1997, 2000]

21
A Appendix: A Sketch of Solutions
A.1 The Basic Model

The following maximization problem must be solved:

R
Max 01 U (Ct )e Æt
dt (19)
subject to
K_ t = F (KtZ; Rt ) Ct
1
St = S0 Rt dt
0
S_ t = Rt
The current value Hamiltonian is dened as
Hc = U (Ct ) + pt ( Rt ) + qt (F (Kt ; Rt ) Ct ) (20)
where pt denotes the shadow price of the resource constraint and qt denotes the shadow
price of capital accumulation. Computing the First Order Conditions yields the following
equations which must be fullled along an optimal path:
p_t = Æpt (21)
U (Ct ) = qt
0 (22)
pt = qt FR (23)
q_t Æqt = qt FK (24)
with FR = @F (@RKtt;Rt) and FK = @F (@K t . Dierentiating (22) with respect to time and
Kt ;Rt )

substituting (24) yields the consumption rate along an optimal path:

C_ t FK Æ
C
=  (25)
t

Dierentiating (23) with respect to time and using (21) and (24) results in
@F 1
F = R
K (26)
@t FR
Substituting FR = f (xt ) xt f 0(xt ) and FK = f 0(xt ) with f (xt ) = F ( KRtt ; 1) and xt = KRtt
in (26) yields the capital-resource ratio along an optimal path
x_t
xt
=  f (xt)xt
(27)

22
A.2 Technology

The following maximization problem must be solved:

R
Max 01 [U (Ct )
t + !t W (Kt ; St )]e Æt
dt (28)
subject to
K_ t = F (KtZ; Rt ) Ct
1
St = S0 Rt dt
0
S_t = Rt
The current value Hamiltonian is
Hc = U (Ct )
t + !t W (Kt ; St ) + pt ( Rt ) + qt (F (Kt ; Rt ) Ct ) (29)
Computing the First Order Conditions yields
p_t = Æpt (30)

t U (Ct ) = qt
0 (31)
pt = !t WS + qt FR (32)
q_t Æqt = = !t WK qt FK (33)
Dierentiating (31) with respect to time and substituting (33) results in
C_ t FK Æ + t (WK U 0 (Ct ))=U 0 (Ct )
Ct
= 
(34)
with t =
t . If Wk = WS = 0 one gets the following consumption rate along an optimal
!t
path
C_ t
Ct
= FK (Æ + t ) (35)
A.3 Backstop Technology

The following maximization problem must be solved

R
Max 01 U (Ct )e Æt dt (36)
subject to
K_ t = F (KtZ; Rt ) Ct g (zt )Rt
1
St = S0 Rt dt
0
S_t = Rt

23
where g(zt)Rt represents the total extraction costs. The current value Hamiltonian is
Hc = U (Ct ) + pt ( Rt ) + qt (F (Kt ; Rt ) Ct g (zt )Rt ) (37)
Computing the First Order Conditions yields
p_t = Æpt + qt g 0 (zt )Rt (38)
= Æptqt g (zt)z_t
0

= Æptqt @g@t(zt )
U 0 (Ct ) = qt (39)
pt = qt (FR g (zt )) (40)
q_t Æqt = qt FK (41)
Dierentiating (40) with respect to time and using (38) and (41) results in
@F 1 F g (z )
FK = R + K t (42)
@t FR FR
Substituting FR = f (xt ) xt f 0(xt ) and FK = f 0(xt ) in (42) results in the following
capital-resource ratio along an optimal path
x_t f (xt ) f 0 (xt ) g (zt )
x
= x
+ x f 00 (x ) x
(43)
t t t t t

A.4 The Amenity Value of a Natural Resource

The following maximization problem must be solved

R
Max 01 U (Ct ; St )e Æt
dt (44)
subject to
K_ t = F (KtZ; Rt ) Ct
1
St = S0 Rt dt
0
S_t = Rt
The current value Hamiltonian is dened as
Hc = U (Ct ; St ) + pt ( Ct ) + qt (F (Kt ; Rt ) Ct ) (45)
Computing the First Order Conditions yields
p_t Æpt = US (46)
UC = qt (47)
pt = qt FR (48)
q_t Æqt = qt FK (49)
24
Dierentiating (47) with respect to time and substituting (49) results in the following
consumption rate along an optimal path:
C_ t FK Æ (UCS =UC )Rt
= Ct 
(50)
Dierentiating (48) with respect to time and using (46) and (49) yields
@F 1 U =U
F = R + S C
K (51)
@t FR FR
Substituting FR = f (xt ) xt f 0 (xt ) and FK= f 0(xt ) in (51) yields the following capital-
resource ratio
x_t f (x ) U =U
xt
=  t + 2 S 00 C
xt xt f (xt )
(52)

A.5 Estimation

The following maximization problem has to be solved:

R1 Ct1 
Max 0 1 
e Æt
dt (53)
subject to (54)
K_ t = Kt Rt1 
Ct (55)
S_t = Rt (56)
The optimality conditions are
C_ t FK Æ
Ct
= 
(57)
x_t f (xt )
xt
= xt
(58)
With FK = Kt 1 R1
t

=  ( RKTt ) 1 = xt 1 and F ( KRtt ; 1) = f (xt ) = x one gets:
C_ t xt 1 Æ
Ct
= 
(59)
x_t
x
= xt 1 (60)
t

25
Setting yt = RCtt yields yy_tt = CC_tt R_t
Rt and the following estimable system is obtained:
y_t xt 1 Æ
yt
= 
 (61)
x_t
xt
= xt 1 (62)
R_ t
Rt
=  (63)
where  denotes the growth rate of the resource ow.

26
B Appendix: Data Sources
 Annual data from 1960 to 1995 for the following time series are, taken from the
U.S. Department of Commerce, Economics and Statistics, Bureau of the Census,
\Statistical Abstract of the United States, 1965 { 1996"
{ crude petroleum production
{ coal production
{ natural gas production
{ copper production
{ iron ore production
{ zinc production
{ lead production
{ share of nuclear power, renewable energy and fuel minerals in total energy
production
{ total mineral production value
{ proved reserves of crude petroleum and natural gas
{ gross private consumption
{ accumulated gross xed capital formation
 Estimated reserves of crude petroleum and natural Gas are from U.S. Geological
Survey, \National Assessment of Oil and Gas Resources, 1995"
{ proved reserves
{ eld growth
{ undiscovered resources
{ total reserves
 Estimated reserves of coal are from Energy Information Administration (1996):
{ proved reserves
{ other reserves
{ total reserves
 Estimated reserves of copper, iron ore, lead and zinc are from US Geological Survey,
\Mineral Commodity Summary, 1996"
{ proved reserves
{ other reserves
{ total reserves
27
References
[1]Aghion, P. and P. Howitt . Endogenous Growth Theory. Cambridge, Mas-
sachusetts, London: The MIT Press (1998).
[2]Barbier, E. B. "The Concept of Sustainable Economic Development." Environ-
mental Conservation (1987), 14 (2): 101-109.
[3]Barbier, E. B. New Approaches in Environmental and Resource Economics. Inter-
national Institute for Environment and Development (1989).
[4]Becker, R. "Intergenerational Equity: The Capital-Environment Trade-O." Jour-
nal of Environmental Economics and Management (1982), 9: 165-185.
[5]Beckerman, W.. In Defence of Economic Growth. London: The Trinity Press
(1974).
[6]Beltratti, A. . "Sustainable Growth: Models and Policy Implications." In Gianna
Boero and Aubrey Silberston, eds., Environmental Economics, Proceedings of a con-
ference held by the Confederation of European Economic Associations at Oxford
(1993), pp. 296-311.
[7]Beltratti, A.. Models of Economic Growth with Environmental Assets. Dordrecht,
Boston, London: Kluwer Academic Publishers (1996).
[8]Beltratti, A. . "Intergenerational Equity and Environmental Preservation." Nota
Di Lavoro (1995), pp. 1-15.
[9]Beltratti, A. andG. Chichilnisky and G. M. Heal. "Sustainable Growth and
the Green Golden Rule." NBER Working Paper Series (1993), No. 4430.
[10]Beltratti, A. andG. Chichilnisky and G. M. Heal. "The Environment and
the Long Run: a Comparison of Dierent Criteria." Richerche Economiche (1994),
48: 319-340.
[11]Chichilnisky, G. . "What is Sustainable Development?" Paper presented at Stan-
ford Institute for Theoretical Economics, 1993. Published as "An Axiomatic Ap-
proach to Sustainable Development." Social Choice and Welfare (1996), 13 (2): 219-
248.
[12]Clark, C. "Mathematical Bioeconomics" New York, J. Wiley (1990)
[13]Constanza, R. "The Ecological Eects of Sustainability: Investing in Natural Cap-
ital." In Robert Goodland, Herman Daly and Salah El Serafy, eds., Environmentally
Sustainable Economic Development Building On Brundtland. Environment Working
Paper No. 46, The World Bank (1991).
28
[14] Daly, H. E. "The Economics of the Steady State." American Economic Association
(1974), 64 (2): 15-21.
[15] Daly, H. E. "The Economic Growth Debate: What Some Economists Have Learned
But Many Have Not." Journal of Environmental Economics and Management (1987),
14 (4): 323-336.
[16] Daly, H. and J. B. Cobb. For the Common Good: Redirecting the Economy to-
ward Community, the Environment and a Sustainable Future. Boston: Beacon Press
(1989).
[17] Dasgupta, P.. "Exhaustible Resources: Resource Depletion, Research and Devel-
opment and the Social Rate of Discount." In Richard Layard, Stephen Glaister,
eds., Cost-Benet Analysis. Cambridge and New York: Cambridge University Press
(1994).
[18] Dasgupta, P. and G. M. Heal . "The Optimal Depletion of Exhaustible Re-
sources." Review of Economic Studies, Symposium on the Economics of Exhaustible
Resource (1974), pp. 3-28.
[19] Dasgupta, P. and G. M. Heal. Economic Theory and Exhaustible Resources.
Welwyn: Cambridge University Press (1979).
[20] Dasgupta, P. andG. M. Heal andM. Majumdar. "Resource Depletion and
Research and Development." In Georey Heal, ed., The Economics of Exhaustible
Resources. International Library of Critical Writings in Economics, Edward Elgar
(1993).
[21] Davison, R. "Optimal Depletion of an Exhaustible Resource with Research and De-
velopment towards an Alternative Technology." Review of Economic Studies (1978),
45 (2): 355-368.
[22] Dixit, A. andP. Hammond andM. Hoel . "On Hartwick's Rule for Regular Max-
imin Paths of Capital Accumulation and Resource Depletion." Review of Economic
Studies (1980), 47 (3): 551-556.
[23] Energy Information Administration . US Crude Oil, Natural Gas, and Natural
Gas Liquids Reserves 1997. Annual Report (1997).
http : ==www:eia:doe:gov=oil gas=natural gas=nat frame:html
[24] Energy Information Administration. Coal Industry Annual 1996. Executive
Summary (1996).
http : ==www:eia:doe:gov=cneaf=coal=cia=cia sum:html:

29
[25] Georgescu-Roegen, N. The Entropy Law and the Economic Process. Cambridge:
Harvard University Press (1971).
[26] Goldsmith, E. "A Blueprint for Survival." The Ecologist (1972), 2 (1): 1-43.
[27] Forrester, J. World Dynamics. Cambridge: Wright Allen (1971).
[28] Hanley, N. andC. L. Spash. Cost-Benet Analysis and the Environment. Chapter
8, Bodmin: Edward Elgar (1993).
[29] Hartwick, J. M. "Intergenerational Equity and the Investing of Rents from Ex-
haustible Resources." Review of Economic Studies (1977), 67 (5): 973-974.
[30] Hartwick, J. M. andN. D. Olewiler . The Economics of Natural Resource Use.
Chapter 6, New York: Harper & Row (1986).
[31] Heal, G. M. "The Relationship between Price and Extraction Cost for a Resource
with a Backstop Technology." In Georey Heal, ed., The Economics of Exhaustible
Resources. International Library of Critical Writings in Economics, Edward Elgar
(1993).
[32] Heal, G. M. "The Optimal Use of Exhaustible Resources." In Alan Kneese and
James Sweeney, eds., Handbook of Natural Resource and Energy Economics. Vol 1.
Amsterdam: North Holland (1985).
[33] Heal, G. M. "Interpreting Sustainability." Nota Di Lavoro (1995), 1: 1-14.
[34] Heal, G. M. Valuing the Future, Economic Theory and Sustainability. Columbia:
Columbia University Press (1998).
[35] Hotelling, H. "The Economics of Exhaustible Resources." Journal of Political
Economy (1931), 39: 137-175.
[36] Kranzberg, M. "Can Technology Provide for the Future?" In Andrew Weintraub,
Eli Schwartz, J. Richard Aronson, eds., The Economic Growth Controversy. New
York: International Arts & Sciences Press (1973).
[37] Krautkraemer, J. A. "Optimal Growth, Resource Amenities and the Preservation
of Natural Environments." Review of Economic Studies (1985), 52: 153-171.
[38] Lucas, R. . "On the Mechanics of Economic Growth." Journal of Monetary Eco-
nomics (1988), 22: 3-42.
[39] Malthus, Th. R. An Essay on the Principle of Population. Oxford: Oxford Uni-
versity Press (1993).
30
[40] Meadows D. H. and D. L. Meadows and J. Randers and W. W. Behrens .
The Limits to Growth. New York: Universe Books (1972).
[41] Meadows D. H. and D. L. Meadows and J. Randers . Beyond the Limits. Post
Mills: Chelsea Green Publishing Company (1992).
[42] Mishan, E. J. "Growth and Antigrowth: What are the Issues?" In Andrew Wein-
traub, Eli Schwartz, J. Richard Aronson, eds., The Economic Growth Controversy.
New York: International Arts & Sciences Press (1973).
[43] Nordhaus, W. D. "World Dynamics: Measurement without Data." The Economic
Journal (1973), 15 (7): 1156-1183.
[44] Nordhaus, W. D. "Lethal Model 2: The Limits to Growth Revisited." Brookings
Papers on Economic Activity (1992), 2: 1-59.
[45] Nordhaus, W. D. "Is Growth Sustainable? Re ections on the Concept of Sustain-
able Economic Growth." Paper presented to the International Economic Association
Conference (1992), Varenna, Italy.
[46] Oren, S. S. and S. G. Powell. "Optimal Supply of a Depletable Resource with
a Backstop Technology: Heal's Theorem Revisited." In Georey Heal, ed., The Eco-
nomics of Exhaustible Resources. International Library of Critical Writings inEco-
nomics, Edward Elgar (1993).
[47] Pearce, D. and A. Markandya andE.B. Barbier. Blueprint for a Green Econ-
omy. London: Earthscan Publications, (1989).
[48] Pearce, D. and A. Markandya andE. B. Barbier. Sustainable Development:
Economics and Environment in the Third World. Worcester: Edward Elgar (1990).
[49] Perman, R. and Y. Ma and J. McGilvray. Natural Resources and Environmental
Economics, London: Longman (1996).
[50] Pezzey, J. Economic Analysis of Sustainable Growth and Sustainable Development.
The World Bank Policy Planning and Research Sta, Environment Department
Working Paper (1989) No. 15.
[51] "The Golden Rule of Accumulation: A Fable for Growthmen." Amer-
Phelps, E. S.
ican Economic Review (1961), pp. 638-643.
[52] Ramsey, F.. "A Mathematical Theory of Saving." Economic Journal (1928), 38:
543-559.
[53] . A Theory of Justice. Oxford, England: Clarendon (1972).
Rawls, J.

31
[54] Redclift, M. Sustainable Development. London and New York: Routledge (1989).
[55] Romer, P. . "The Origins of Endogenous Growth." Journal of Economic Perspectives
(1994), 8 (1): 3-22.
[56] Semmler, W. and G. Gong , "Estimating Parameters of Real Business Cycle Mod-
els", Journal of Economic Behaviour and Organization, vol 30 (1997): 301 { 325.
[57] Semmler, W. and Sieveking, M. "On the Optimal Exploitation of Interacting
Resources." Journal of Economics (1994), vol. 59(1).
[58] Semmler, W. and Sieveking, M. "Debt Dynamics and Sustainable Debt." Journal
of Economic Dynamics and Control, 24 (2000): 1121 { 1144.
[59] Sieveking, M. and Semmler, W. "The Present Value of Resources with Large
Discount Rates." Applied Mathematics and Optimization (1997), vol. 38:283-309
[60] Silberberg, E. The Structure of Economics: A Mathematical Analysis. New York:
McGraw-Hill Book Company (1978).
[61] Solow, R. M. "Is the End of the World at Hand?" In Andrew Weintraub, Eli
Schwartz, J. Richard Aronson, eds., The Economic Growth Controversy. New York:
International Arts & Sciences Press (1973).
[62] Solow, R. M. "Intergenerational Equity and Exhaustible Resources." Review of
Economic Studies, Symposium on the Economics of Exhaustible Resources (1974),
pp. 29-45.
[63] Solow, R. M. "Sustainability: An Economist's Perspective." Eighteenth J. Seward
Johnson Lecture (1991), Marine Policy Center.
[64] Solow, R. M. "An Almost Practical Step toward Sustainability." An Invited Lecture
on the Occasion of the Fourtieth Anniversary of Resources for the Future (1992).
[65] Smulders, S. "Entropy, Environment, and Endogenous Economic Growth." Inter-
national Tax and Public Finance (1995), 2: 319-340.
[66] Stiglitz, J.E. "Growth with Exhaustible Natural Resources: EÆcient and Opti-
mal Growth Paths." Review of Economic Studies, Symposium on the Economics of
Exhaustible Resources (1974), pp. 123-137.
[67] Stiglitz, J. E. "A Neoclassical Analysis of the Economics of Natural Resources."
In Georey Heal, ed., The Economics of Exhaustible Resources. International Library
of Critical Writings in Economics, Edward Elgar (1993).

32
[68] Toman, M. A. and J. Pezzeyand J. A. Krautkraemer. "Neoclassical Eco-
nomic Growth Theory and Sustainability." In Daniel W. Bromley, ed., Handbook of
Environmental Economics, Oxford UK & Cambridge USA: Blackwell Press (1993).
[69] US Department of Commerce, Economics and Statistics, Bureau of the
Cencus. Statistical Abstract of the United States. 1965 - 1997.
[70] US Geological Survey. A Summary of the US Geological Survey 1995.
http : ==energy:usgs:gov=factsheets=NAresults=nat:assess:number:html:
[71] US Geological Survey. National Assessment of Oil and Gas Resources. (1995).
http : ==energy:usgs:gov=products=factsheets=F S 024 98=index:html:
[72] US Geological Survey. Minerals Information 1997.
http : ==minerals:usgs:gov:
[73] US Geological Survey. Mineral Commodity Summary 1999.
http : ==minerals:usgs:gov:
[74]  cker, C.C.
Weiza "Lemmas for a Theory of Approximate Optimal Growth." Review
of Economic Studies (1967), pp. 143-151.
[75] World Commission on Environment and Development. Our Common Fu-
ture (the Brundtland Report). New York: Oxford University Press (1987).

33